FlexLLama v0.1.8 Release Notes
Overview
This release brings official Docker support for Vulkan GPUs, Windows Docker improvements, and Python 3.10 compatibility. It also includes a fix for the auto-unload mechanism ensuring reliable resource management.
🚀 New Features
Docker Support for Vulkan GPUs
Added: Comprehensive Docker support for Vulkan-capable GPUs (Intel Arc/iGPU, AMD, NVIDIA). Includes a new Dockerfile.vulkan, docker-compose.yml updates, and enhanced startup scripts (docker-start.sh, docker-start.ps1).
Impact: Users can now run hardware-accelerated inference in Docker on a wider range of consumer hardware using Vulkan, expanding deployment options beyond CUDA.
Python 3.10 Support
Added: Verified compatibility and support for Python 3.10.
Impact: Broader compatibility with modern Python environments and libraries.
🔧 Bug Fixes
Docker on Windows
Fixed: Resolved line ending issues and Dockerfile compatibility specific to Windows environments.
Impact: Smoother and more reliable Docker deployment experience for Windows users.
Unlock the power of Vulkan in Docker with FlexLLama v0.1.8!